AI-Based Early Detection and Stage Classification of Parkinson's Disease Using Machine Learning

Repeated voice measurements offer a low-cost way to study Parkinson’s disease (PD), but their interpretation depends on how recordings are grouped and which clinical labels are available. We present a two-task design for PD screening and ordinal stage estimation. For a preliminary screening experiment, we averaged 22 acoustic features within each recording identifier group in the UCI Parkinsons file. A nearest-centroid classifier correctly classified 24 of 32 held-out groups, including 17 of 24 PD-labelled groups and seven of eight control groups. An always-positive rule obtained the same accuracy while identifying no controls. These findings motivate separate reporting of sensitivity and specificity. Clinical stage estimation remains a proposed extension requiring independently recorded stage labels.

Authors

Institutions

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-11-19
DOI
https://doi.org/10.5281/zenodo.24375666
Primary Topic
Voice and Speech Disorders
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

AI-Based Early Detection and Stage Classification of Parkinson's Disease Using Machine Learning

Yashvardhan Rajvardhan Dongare, Vishishta Anil Prasad, Mohammad Aman Khan
Zenodo (CERN European Organization for Nuclear Research)
Voice and Speech Disorders
preprint

AI-Based Early Detection and Stage Classification of Parkinson's Disease Using Machine Learning

Yashvardhan Rajvardhan Dongare, Vishishta Anil Prasad, Mohammad Aman Khan
preprint en

Abstract

Repeated voice measurements offer a low-cost way to study Parkinson’s disease (PD), but their interpretation depends on how recordings are grouped and which clinical labels are available. We present a two-task design for PD screening and ordinal stage estimation. For a preliminary screening experiment, we averaged 22 acoustic features within each recording identifier group in the UCI Parkinsons file. A nearest-centroid classifier correctly classified 24 of 32 held-out groups, including 17 of 24 PD-labelled groups and seven of eight control groups. An always-positive rule obtained the same accuracy while identifying no controls. These findings motivate separate reporting of sensitivity and specificity. Clinical stage estimation remains a proposed extension requiring independently recorded stage labels.

Zenodo (CERN European Organization for Nuclear Research)
MIT Art, Design and Technology University (IN)
Voice and Speech Disorders
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon